The chicken cornea as a model of wound healing and neuronal re-innervation

Animals

Animals were used in accordance with international standards for animal treatment established by the National Institutes of Health, ARVO and The Ohio State University. Newly hatched white leghorn chickens (Gallus gallus domesticus) were obtained from the Ohio State University Department of Animal Sciences (Columbus, OH) and raised on a cycle of 12 h light, 12 h dark (light 7 AM to 7 PM) in a stainless steel brooder. Chicks were fed Purina chick starter (Purina, St Louis, MO) and water ad libitum.

Intraocular injections and corneal wound generation

Animals were anesthetized by inhalation of 2.5% isoflurane in O₂ at a flow rate of 1.5 l/min. Corneal anesthesia was achieved using 1 drop of 0.5% topical proparacaine ophthalmic solution. A 4 mm Barraquer pediatric lid speculum was inserted and the chick placed under a Leica S6E stereo microscope (Leica, Buffalo, NY). A single linear corneal incision was made, nasal to temporal, using a Nichamin LRI (Storz, Rochester NY) diamond knife set for a blade depth of 0.50 mm through the central cornea with a mid-stroma target depth. After incision, 1 drop of 0.3% moxifloxacin topical ophthalmic solution (Vigamox; Alcon, Fort Worth, TX) was applied and animals were subsequently monitored for signs of corneal infection.

Cellular proliferation was assayed by using intravitreal injections of the thymidine analog 5-bromo-2-deoxyuridine (BrdU). Six μg of BrdU diluted in 20 μl sterile saline was injected into the dorsal quadrant of each eye through the superior eyelid.

Tissue dissection, fixation, and sectioning

Animals were sacrificed; the eyes were enucleated and hemisected along the equator of the eye. Anterior segments were fixed in 4% paraformaldehyde (PFA) + 3% sucrose in 0.1 M PB, pH 7.4 for 25 min at room temperature. Samples were washed twice for 15 min in 0.05M phosphate buffered saline (PBS) and cryoprotected in 20% sucrose in PBS overnight. Anterior segments were placed in Optimal Cutting Temperature (OCT) compound (Sakura Tissue Tek, Torrance, CA) medium and freeze-mounted onto sectioning blocks. Corneal sections, nominally 12 microns thick, were made transversely through the corneal wound using a Microm HR550 cryostat (Thermo Fisher, Walldorf, Germany). Sections were thaw-mounted on SuperFrost Plus slides (Thermo Fisher, Fairlawn, NJ) and stored at –20 °C until immunolabeling.

Corneal whole-mounts were obtained after the anterior segment was fixed in 4% PFA as described above. Corneas were dissected away from anterior segments. Isolated corneas were equilibrated in 20% sucrose in PBS and taken through 3 cycles of freezing and thawing. Cornea whole-mounts were permeabilized using graded washes of DMSO (40%→ 60%→ 80%→ 100%→ 80%→ 60%→ 40%). Individual corneas were placed in 400 μl of primary antibody solution (antiserum + PBS + 0.2% Triton X-100 + 0.01% NaN₃) and placed on a nutating shaker overnight. The primary antibody solution was aspirated and corneas were washed two times in PBS. Corneas were covered with 400 μl of secondary antibody solution (Alexa Fluor [Invitrogen, Carlsbad, CA] antiserum + PBS + 0.2% Triton X-100 + 0.01% NaN₃) and placed on a nutating shaker overnight. Slides were mounted under cover glass on 80% glycerol in water.

Immunolabeling and photography

Slides were warmed to room temperature and ringed with rubber cement. Sections were washed two times in 0.05 M PBS for 10 min and covered with 250 μl of primary antibody solution (antiserum + PBS + 0.2% Triton X-100 + 0.01% NaN₃ + 5% normal goat serum or normal donkey serum) and incubated for 24 h in a covered, humidified chamber at room temperature. The primary antibodies included: (1) mouse anti-BrdU used at 1:80 dilution (G3G4; Developmental Studies Hybridoma Bank [DSHB]; University of Iowa, Iowa City, IA), (2) mouse anti-CD45 used at 1:200 dilution (HIS-C7; Cedi Diagnostics, Prionics AG, Zurich, Switzerland), (3) mouse anti-α-smooth muscle actin (αSMA) used at 1:500 dilution (1A4; Sigma-Aldrich, St Louis, MO), (4) rat anti-substance P used at 1:400 dilution (Ab6338; Abcam, Cambridge, MA), (5) mouse anti-synaptic vesicles used at 1:1,000 dilution (SV2; DSHB), (6) mouse anti-vimentin used at 1:80 dilution (H5; DSHB), (7) mouse anti-vinculin used at 1:50 dilution dilution (VN3–24; DSHB), (8) mouse anti-perlecan used at 1:50 dilution (5C9; DSHB), and (9) rabbit anti-cleaved caspase 3 (CC3) used at 1:1,200 dilution (AF835; R&D Systems, Minneapolis, MN). The primary antibody solution was aspirated and slides were washed in two times in 0.05 M PBS for 10 min. Slides were covered with 250 μl of secondary antibody solution (antiserum + PBS + 0.2% Triton X-100 + 0.01% NaN₃) in a covered, humidified chamber at room temperature for 1 h. Slides were washed in PBS and cover glass mounted on 4:1 glycerol to water (volume:volume). Antigen retrieval was used to permit immunolabeling for BrdU; sections were washed for 7 min in 4 M HCl. Secondary antibodies included donkey-anti-goat-Alexa488, goat-anti-rabbit-Alexa488, goat-anti-mouse-Alexa488/568, rabbit-anti-goat Alexa488 and goat-anti-mouse-IgM-Alexa568 (Invitrogen) diluted to 1:1,000 in PBS plus 0.2% Triton X-100.

We evaluated the specificity of primary antibodies by comparison with published examples of results and assays for specificity. None of the observed labeling was due to non-specific labeling of secondary antibodies or autofluorescence because sections labeled with secondary antibodies alone were devoid of fluorescence. Secondary antibodies included goat-anti-rabbit-Alexa488/568 and goat-anti-mouse-Alexa488/568 (Invitrogen) diluted to 1:1000 in PBS plus 0.2% Triton X-100.

Labeling of dying cells

Dying cells were identified using Terminal deoxynucleotidyl Transferase Biotin-dUTP Nick End (TUNEL) Labeling (In situ Cell Death Kit TMR; Roche, Indianapolis, IN) according to the protocols provided by the manufacturer. To label cells with compromised plasma membranes we applied a single topical drop (50 µl; 0.2 mM in saline) of Ethidium Homodimer-1 (EtHD; Live/Dead® Cell Viability Assay; Invitrogen) shortly (<15 min) after incisions were made into the cornea. Fifteen min after the application of the EtHD the corneas were harvested and fixed, as described above.

Microscopy, cell counts and statistics

Photos were taken using a Leica DM 5000B microscope and Leica DC500 12-megapixel cooled CCD camera. Confocal microscopy was performed using a Zeiss LSM 510 system (Zeiss, Thornwood, NY) at the Hunt-Curtis Imaging Facility at The Ohio State University. Images were optimized for color, brightness and contrast, and double-labeled images overlaid by using Adobe Photoshop^™6.0 (Adobe, San Jose, CA). Cells counts were performed and statistical analysis completed using SPSS V17.0 (IBM, Inc, Armonk, NY).

Cellular proliferation after corneal insult

To identify newly generated cells in the cornea we injected BrdU into the vitreous chamber and harvested corneas at different times after BrdU-treatment. Corneas were harvested (and BrdU injected) at day 1 (BrdU injected 4 h prior), day 4 (BrdU injected at days 1, 2, and 3), day 7 (BrdU injected at days 4 and 5), and day 14 (BrdU injected at days 7 and 8). Under control conditions with no incision, BrdU-positive cells were observed near the basement membrane of the corneal epithelium (Figure 1A-C), consistent with the notion that mitotically active cells are normally found in basal epithelial layers. In addition, we occasionally observed BrdU-positive cells in the corneal stroma, consistent with the quiescent nature of keratocytes in the normal cornea (Figure 1A-C). After damage, proliferating and newly generated cells were observed in the stroma and epithelium immediately surrounding the corneal incision. One day after corneal incision, the wound was filled by a layer of epithelial cells that were negative for BrdU (Figure 1D). By comparison, BrdU-positive cells were observed at the ridge of the corneal epithelium leading into the trough caused by the incision (Figure 1D). Few BrdU-positive cells were observed in the corneal stroma immediately below the epithelium (Figure 1D). At 4 days after injury, there were no BrdU-positive cells in the epithelium overlying the corneal incision (Figure 1E). By comparison, there was a significant increase in BrdU-positive cells observed within an area of hypercellularity in the anterior stroma immediately underlying the site of incision (Figure 1E,H). At 7 days after injury, the number of BrdU-positive cells within the zone of injury decreased compared to that seen at 4 days after injury (Figure 1F,H). At 14 days after injury, many BrdU-positive cells were observed in basal layers of the epithelium in the region of the corneal incision (Figure 1G). In addition, at day 14, we observed a few BrdU-positive cells in the stroma at the edges of the incision site (Figure 1G,H).

Counts of Draq5-labeled nuclei in the corneal stroma were performed to assess the accumulation of cells following injury. In control corneas, the average density was 74.70 nuclei per 20,000 µm² (n=20, SD±12.97). There were significant increases in cell density in the stroma after injury (Figure 1A-G,I). At 1 day after injury, the cell density within the wound-region of the stroma was not significantly different from that of control corneas (Figure 1D,I). At 4 days after injury, there was a significant, approximate twofold increase in cellular density in the stroma within the zone of injury (Figure 1E-I). The increased cell density that was seen at 4 days after injury was not significantly different at 7 and 14 days after injury (Figure 1F-G,I).

It is possible that the increased cell density of stromal cells in the corneal wound occurred due to cellular migration. Thus, we examined whether there was a depletion of cell density in the stromal regions immediately adjacent to the corneal injury. There was no significant difference in stromal cell density in uninjured corneas compared to the region adjacent to the injury for any time point (Figure 1I). Moreover, there was no significant change in cell density in the stromal region adjacent to the scar, at any time point after injury (Figure 1I).

Expression of filamentous proteins after corneal insult

Following corneal incision, we found an upregulation of the intermediate filament vimentin in stromal cells. Control corneas had vimentin-immunoreactivity in most, if not all, stromal keratocytes (Figure 2A), similar to previous descriptions [17,18]. Vimentin was not observed in the corneal epithelial cells, whereas vimentin was observed in the endothelial cells (Figure 2A). At 1 day after corneal incision, vimentin appeared modestly upregulated in stromal cells directly under the corneal epithelium where Bowman’s membrane was disrupted (Figure 2B). At 4 days after corneal incision, vimentin-immunofluoresence was dramatically increased in stromal cells within the wound, and these cells extended several hundred micrometers into deeper layers of the corneal stroma (Figure 2C). In the stromal areas lateral to the site of incision, Bowman’s membrane appeared to remain intact and the expression of vimentin was relatively low, similar to levels observed in control corneas (Figure 2C). At 7 days after injury, levels of vimentin-immunoreactivity remained elevated, similar to levels observed at 4 days after injury (Figure 2C,D). Levels of vimentin-immunoreactivity remained elevated in stromal cells at 14 days after treatment, but the stromal cells that expressed high levels of vimentin appeared to spread laterally away from the site of incision (Figure 2E). Vimentin was not observed in the corneal epithelium at any time point after corneal incision.

We next examined the expression of vinculin, a membrane-associated cytoskeletal protein, in damaged corneas. In control corneas, vinculin was expressed at low levels in stromal cells, whereas vinculin was not detected in the epithelium (Figure 2F) or in the endothelium (not shown). One day after injury, vinculin expression in the stromal cells remained unchanged (Figure 2G). There was no vinculin-immunofluorescence observed in the regenerated epithelium overlying the corneal wound at any time point after injury (Figure 2G-J). At 4 days after injury, there was a dramatic increase in the expression of vinculin in stromal cells within the region of corneal incision (Figure 2H). Vinculin expression remained elevated in the wound-activated stromal cells through 7 days after injury (Figure 2I). By 14 days after injury, the expression of vinculin was decreased compared to expression levels seen at 7 days after injury (Figure 2I,J). However, levels of vinculin expression remained elevated at 14 days after injury compared to levels seen in controls, and appeared to spread laterally away from the site of incision compared to the distribution of reactive cells seen at earlier time points (Figure 2J).

We next probed for the expression of perlecan, a heparan sulfate proteoglycan that is associated with the endothelial basement membrane in the cornea [19,20]. In untreated corneas, perlecan was observed in the posterior cornea near Descemet’s membrane (Figure 3A) and at high levels in the endothelium (data not shown), consistent with previous reports [21,22]. Perlecan-immunofluorescence was detected at relatively low levels in the anterior stroma, and was not detected in Bowman’s membrane or the epithelium (Figure 3A). At 4 h and 1 day after injury, perlecan expression in the stroma was unchanged compared to that seen in control corneas (Figure 3A-C). At 4 days injury, perlecan expression was upregulated in stromal cells within the site of incision, corresponding to the area of increased cell density (Figure 3D). At 7 days after injury, perlecan remained upregulated in stromal cells in the area of the wound (Figure 3E). This focal upregulation of perlecan was observed at 14 days after, but appeared to spread laterally away from the site of incision (Figure 3F). We failed to detect perlecan in the corneal epithelium, which is not consistent with reports of perlecan in the cornea of chicks and other species. The 5C9 monoclonal antibody to perlecan recognizes domain IV. It is possible that a splice variant of perlecan that excludes domain IV is present in the epithelium of the chick cornea, and, therefore, was not detected in our studies.

Antibodies to alpha-smooth muscle actin (αSMA) were used to examine injury-induced changes in keratocytes after corneal injury. In control chick corneas, αSMA was not observed (Figure 3G). At 1 day after injury, αSMA was not detectable in stromal cells (not shown). By contrast, 4 days after injury αSMA was markedly upregulated in the stromal cells immediately under the area of incision (Figure 3H,I). By using confocal microscopy and 3D reconstruction, we found αSMA surrounding the nuclei of activated corneal fibroblast-like cells (Figure 3I). αSMA-expressing stromal cells were present immediately posterior to Bowman’s membrane and extended approximately 200 µm into the stroma (Figure 3H-I). Increased levels of α-SMA-immunofluoresence continued through 7 and 14 days after injury (not shown).

Invasion of immune cells into injured corneas

Activated macrophages are known to accumulate and participate in the wound healing response of the stroma in damaged corneas [23] (reviewed by [4]). To identify monocytes that invade the cornea we applied antibodies to CD45. In undamaged corneas, we identified numerous CD45+ cells scattered across the limbal region of the cornea (Figure 4A). There were few CD45+ cells in the stroma of undamaged corneas (Figure 4B). At 1 day after injury, there was a significant accumulation of CD45+ cells in anterior regions of the stroma, surrounding the site of incision (Figure 4C). The CD45+ cells appeared to be reactive macrophages with high levels of CD45-expression and ameboid morphophology (Figure 4C,G) At 4 days after injury, there were many CD45+ cells remaining within the stroma at the site of injury (Figure 4D) and scattered throughout the stroma in regions adjacent to the site of incision, across the entire cornea (Figure 4E). Within the site of injury at 7 and 14 days after incision, the abundance of CD45+ cells appeared diminished compared to numbers seen at day 4 (Figure 4F,H). At 14 days after injury, the abundance of CD45+ cells in the stroma outside of the injured region was reduced to levels seen in undamaged corneas (Figure 4B,I). To determine whether any of the CD45+ monocytes that accumulate in the stroma are derived through proliferation, we probed for BrdU-labeling at 1 (BrdU injected 4 h before harvest) and 4 days (BrdU injected at days 1, 2, and 3). We found very few of the CD45+ positive monocytes that accumulate within the stroma in response to injury were labeled for BrdU (Figure 4J,K). At 4 days after incision, by comparison, we found that none of the BrdU-labeled cells that accumulate within the injury zone of the stroma were CD45+ monocytes (Figure 4L).

Corneal innervation after corneal incision

Branches and terminals of the ophthalmic branch of the trigeminal nerve (Cranial Nerve V₁) in the cornea were labeled using antibodies to SV2, a synaptic vesicle protein. Examination of control corneas revealed extensive neuronal processes running in the basal layers of the epithelium in peripheral and central regions of the cornea (Figure 5A,B). The SV2-labeled neuronal processes appeared to run in bundles of axons in deeper layers of the stroma near the limbus (Figure 5A). The axons in the stroma appeared to ascend and terminate with numerous dense endings in basal layers of the epithelium (Figure 5A,B). At 1 day after corneal incision, few neuronal processes were observed in the basement membrane and epithelium overlying the wound (Figure 5C). There were neuronal processes at the edge of the site of incision in the regenerated epithelium, whereas there were no processes in the stroma immediately underlying the injury (Figure 5C). At 4 days after injury, numerous SV2-positive endings were present in the regenerated epithelium at the edge of the incision and within the trough of the incision (Figure 5D-J). At 7 days after injury, SV2-positive processes were observed in the epithelium overlying the site of injury (Figure 5K). At 14 days after injury, SV2-positive neuronal processes, that were presumably regenerated, were observed within the injured stromal region where the cell density was elevated (Figure 5L).

The neuropeptide substance P has been reported to be released from the axon terminals that innervate the cornea, and this peptide is known to regulate the proliferation and turn-over of corneal epithelial cells (reviewed by [24]). Thus, we probed for substance P in axons and nerve terminals that re-innervate damaged regions of cornea. We found substance P-immunoreactivity in all SV2-positive axons in the stroma and terminals in the epithelium in undamaged corneas (data not shown). In addition, we found substance P-immunoreactivity in the SV2-positive terminals in the regenerated epithelium that filled the injury site at 4 days after incision (Figure 5M-O).

Cell death after corneal injury

Examination of undamaged corneas revealed no TUNEL-positive cells in the stroma, endothelium or epithelium (Figure 6A). After injury, TUNEL-positive cells were observed in the corneal epithelium and within the stroma. At 2 h after injury, numerous TUNEL-positive cells were observed lining the border of the corneal incision, within the epithelium and stroma that flanked the site of incision (Figure 6B). Numbers of TUNEL-labeled cells appeared increased at 4 h after injury (Figure 6C). At 1 day after injury, few TUNEL-positive cells were observed in the regenerated epithelium that filled-in the site of incision, and no TUNEL-positive cells were observed within the underlying stroma (Figure 6D). Four days after injury, we observed TUNEL-positive cells in the corneal epithelium overlying the wound (Figure 6E). Most of the TUNEL-positive cells were found in the superficial layers of the epithelium overlying the site of injury (Figure 6E). Very few TUNEL-positive cells were observed in the deep layers of the stroma within the region of the incision (Figure 6E). At 7 days after injury, TUNEL-positive cells were not observed in the epithelium or the stroma in damaged corneas (Figure 6F).

To compliment the data from the TUNEL studies, we applied ethidium homodimer (EtHD) shortly (~15 min) after injury. In undamaged corneas, EtHD does not label any cells within the epithelium, stroma or endothelium (Figure 6H-J). By contrast, we identified numerous dead or dying epithelial cells, labeled with EtHD, immediately surrounding the site of injury (Figure 6G,K-N). In addition, we found a few EtHD-labeled cells within the stroma at the edge of the incision site and extending as far as 40 µm into the stroma (Figure 6K-M,O). Labeling with EtHD suggests that epithelial cells and stromal cells die of necrosis shortly after injury. We failed to find apoptotic cells immunolabeled for cleaved caspase 3 (CC3) within injured corneas at any time after damage (data not shown).